Fetching the paper…

Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models · Around